Sonya Huang and Sarah Guo on what's next in AI | Mercury Spheres

Foreign.

Super excited about this panel. We have Sonia Huang from Sequoia Capital. Sonia recently invested in Mercury, so obviously she's most excited about us. But she's also a prolific investor in AI. She did OpenAI, LangChain, Fireworks, and lots of others. And we're going to spend this panel talking about AI mostly. And then we have Sarah Gar. She is the founder of Conviction. Before that she was a Greylock. She's an early stage investor and has backed Runway, Cresta Dust and lots of other fun companies. Also, she's a co host of a podcast called no Priors. So just to kick it off, can each of you share an AI investment you made recently and what got you excited about it?

Sure, happy to. Hi, everyone, I'm Sonia. Let's see. An AI investment I made recently is a company called LangChain. They help companies. They're a developer platform for building LLM applications. And so if you think about what makes large language models very different, they're stochastic in nature instead of code in, code out. Oftentimes it's prompts that are driving these. And as the agents that people are building grow increasingly complex, you actually need really robust orchestration mechanisms for doing things like if the agents can be long running and working in the background and coordinating with networks of agents. And so LangChain was founded by a guy, Harrison Chase, who's been, I think, earlier to the agent trend than anybody. We've been following him since way, way pre chatgpt and just following his view on how agents would shape up. And he's very much building the scaffolding and developer infrastructure to support, support that world. So, yeah, we invested a couple years ago now actually.

Sarah, you're also an investor in LangChain.

I am, actually.

You had a fun story about how you got to be an investor there, right?

Oh, yes. Goodness. So I'm not a very good engineer, but one of the last times I wrote code somewhat involuntarily was I think it was a reasonable expectation for any entrepreneur to be like you. Use my product if you want to talk to me, investor. And actually, speaking of financial visibility, I built a small LangChain application which was essentially like a talk to your finances app, where I wanted to get payroll data and cloud consumption data and just make it easier for early stage founders in our portfolio to understand what was changing costs month to month. But you know, better Mercury than me doing this work. But yes, it was, it was a fun little app.

I think it's a fun story because as entrepreneurs we Often like trying to get investors. But it goes to show you that investors, when they really want you, they'll go write an app for you.

They will.

He made it a. I don't know if Harrison remembers this story but he made it a barrier to taking a meeting. And I was like, okay, we can do this.

Obviously that company was already hot. Any to go back to the question, any interesting AI investments you've made recently?

Yeah, well one company that I'm excited about that we haven't talked much about publicly is this company called Sola. And so I think there's a lot for. I'm sure there are technical and non technical entrepreneurs in this audience. There's a lot for developers to do with language models. There's also many processes in the world that feel very manual and unscalable, that non technical people have the domain understanding for how to do right. And that could be the middle office in a financial services firm. It could be operations in a E commerce company. But what Sola does is give you kind of like a. It looks like a drag and drop like flowchart of processes where you plug in language models to do actions in software that were very manual in your business operations. And so I get really excited when other people get really excited about AI. And so I was doing a customer reference for this company, Solar recently and it was a guy who runs a billion dollar revenue logistics company like a trucking brokerage company. And he's like, we are transforming our business this way. We have 75 workflows developed and there's no engineers in this company. Right. They have like one director of IT person. And so I'm really excited when you are making this stuff more accessible to people. People.

Yeah, I guess who in the audience who's an entrepreneur thinks of their company as an AI first company? Maybe put up your hands just so you guys answer. A smaller ratio than I would have thought, I guess like yeah, as you think about building an AI startup, it feels like the models are doing more and more. Where do you think a founder can operate safely? And like interesting for you both as VCs to invest.

I'll start because we've actually opted at our thinking here. So I would say if you had asked me two years ago, I would have said anything in the application layer like I believe that you're safe. And our belief back then was that the foundation models were going to be so focused on research and building models in a lab that they just didn't have the time and the energy and the expertise to really focus on how do they make their models great for any specific end customer use case. And so I would say overall, like two years ago, my bias to that would have been as long as you're building in the application layer, you're safe. I think that mental model has changed a lot, in part because of just seeing the success that ChatGPT for example, has had. I think a lot of the application layer is actually going to be owned by the foundation models going up the stack. And as a result, I think that if you're building anything kind of horizontal consumer enterprise, I think that's right in the strike zone for the foundation models and where they want to go. I think coding is also right in the strike zone for something they view as a critical path, a critical node to their missions. And so those categories I'd say are relatively tougher to win in. Never say never. But then when it goes to. Sarah mentioned a few of these use cases, we see this all the time of health care. You're a hospital, you really, really, really want just great medical transcription and billing. That's just a different use case than the foundation models are building generally. So I think if you're still building very vertical, specific, domain driven companies, that's pretty probably where we're most interested in investing, where we think those companies are safest, so to speak, from the strike zone of where the foundation models are playing.

Yeah. Sarah, do you have something to add to?

Yeah, I was going to say I feel like the last mile into every use case in the world and all these different professions, all these different industries is actually very long. Right. There are many things that are like earned expertise and trust that are not just intelligence. Right. Like, I'm like, all right, a lot of entrepreneurs with high iq and yet if you go into your other's domains, there's no guarantee you're like that successful. Day one. I will never know as much about financial services innards as IMAD does. And that feels like very important to the success of Mercury at this point. I'm sure it's kind of painful to climb that hill of knowledge.

No one wants to know as much of it. Yeah.

But therein is kind of the advantage. Right. And so I think especially in spaces that are operationally complex, that have regulatory concerns or where honestly the audiences, like they are rightfully so, a little bit skeptical and like pretty concerned about the quality of the knowledge work. Let's say Sequoia and Conviction. Our firms were both invested in this company called Harvey that serves lawyers and a company called Open Evidence which is made medical search for doctors. Right. I use ChatGPT to educate myself about like what, like my kid's snot is purple today. What should I do as like a first line defense? Because it's hard to get ahold of a medical professional. But you know, if somebody is sort of trying to understand the most up to date research for a complex oncology case, I don't really want that doctor like asking ChatGPT. I want them like looking up intelligently research and work by other doctors and medical professionals and that's what they want to. Right. And so I think that there are domains where the specific data you're surfacing and the expertise and the trust just feels very different from some of the consumer applications we'll all use.

We were talking about this earlier. What is your general thoughts on AI and robotics? Is that something that you'd invest in or how are you thinking that will play out?

Yeah, Sonia was talking about how her thinking had changed on you know, the model layer over the last two years is very dangerous to make any like long term claims. And I, I think there's a, there's like a large chance I look very stupid about this a few years from now. But for a decade at my prior firm I would you know, feel very high and mighty on my like practical investing high horse saying like oh partners like robotics is a great way to lose a lot of money over a long period of time. Have you noticed that like China's very good at the supply chain for electronics and robotics and also we have no winning American robotics companies today. And that has been true. I think at risk of being the person that says, you know, this time it's different. I do think the idea of general, you know, general physical intelligence, the ability to like pick up like the task might be something like pick up any cup on your counter, put it in the dishwasher, no matter the shape of the cup and you know, the model of the dishwasher, we can all do that. The robots to date have not been able to. I think this type of generalization we should see in the next few years. It's a question of how expensive the data is. And if that's true, that's very exciting. Then there's just a question of how expensive is the hardware and in what form factors does it get distributed to take a lot of the toil out of the physical work everybody encounters every day, both in their own experiences. I spend, there's only 16, 18 waking hours a day. I spend one hour of every day Chasing my kids around, being like, you know, Lego injury, like dirty clothes, like, we, I do think we will not be doing that 10 years from now. So that is, that is a change in thinking. I don't know how people who have claimed this repeatedly over the last decade have been wrong.

Sonia, what is your take on robots?

I'm the person that's still on my high horse about this one.

That's fair, that's fair.

I will say enough smart people like Sarah are telling me I'm wrong and this time is different, that I'm actually re examining my priorities right now. I've gone and met some humanoid robots last week and all that. I will say I think the future is coming at us a lot faster than I thought it would. Something Sam Altman said at a dinner years ago stuck in my head of robotics. I think the quote was, robotics is hard for the wrong reasons. It's hard because attendance snaps on your robot. It's not hard because you're getting closer to digital intelligence. And that really stuck with me. I think robotics is really, really, really hard. And after 10 years of extremely hard work, you might have something that is good at a small number of household tasks. And so I think the roi, I see it as lower than in the digital realm right now. There's very smart people that will argue with me vehemently on that and I'm very open to be convinced otherwise. I do think my mental model is robotics will play out very similar to autonomous vehicles right now. And I remember when I joined Sequoia, this is eight years ago at this point, there were a lot of, you know, funded self driving companies. And at the end of the day, it took a lot of money, a lot of time. And I think Waymo has made it, Tesla has made it, and there's probably a couple others that are on the brink of making it. But there's a lot of companies that went sideways because it was simply, it was such a hard problem. So for me it's like, it's such a, such a, such a hard problem. But then for the small number of companies that are able to cross that chasm and make it, who here is taking awaymo? It's the most magical thing ever. It's magical. My toddler's taking awaymo. He doesn't even want to ride in Ubers anymore. He just takes, he's like, where's the robot Waymo? And so I do think, like, there's a potential to like, once you get it right in robotics, it's magical. And the Markets are massive. I'm just like, my mental model is like it's going to be like AV and there's a lot of companies that aren't going to make it and don't have the funding to even cross that chasm.

Yeah, I feel like this question might be interesting to the audience since it's not that many AI first entrepreneurs. There's definitely a lot of companies I've invested in that really struggle and they're not an AI company to get attention from VCs. Even if they have like, you know, numbers that a few years ago would have seemed like series A fundable numbers but like, you know, no one cares anymore. It's probably unfair question since both of you do focus on AI. But yeah, how do you advise founders that you know, don't have AI? Maybe they do have like reasonable numbers, but when they're going to investors, how do they get them interested?

I would start by saying we are focused on AI because we are a handful of people and I think it's the most important thing happening in technology today. And that will be true for many years. That being said, I'm not a researcher. I think I'm very interested in solving customer problems and user problems. And so if somebody has a step, function, better solution or business banking or whatever the problem is and the answer is not AI and it is working awesome. We are definitely curious about it. I think the dynamic that Ahmad mentions is real, that some of the things that I can do are like Waymo style magical. Right? She just said it's amazing. And my experience has been like from a opportunity cost perspective and maybe for every entrepreneur here to think about, you know, we want to do the things that are going to have the most impact in the long run. And we think that these magical experiences are one of the things that create opportunities for startups to become really big companies. They're just most different. Right. And so I do think the bar has been raised and it's a combination of digital distribution, which lots of companies benefit from. Right. You can find a better banking solution online now in a more connected world and like magical capabilities. And so companies are getting to revenue targets with their magical products much faster. And I think that is the sort of competitive reality for capital. That being said, there's a lot of capital in the world. I have been very surprised by this over the last few years. It does tend to aggregate toward like a certain set of hot sectors. But I think just like we look at the entrepreneurial ecosystem as not homogenous, like the capital Ecosystem is not homogenous too. There's like you have to find the right investor for you.

I remember when we listened when we met you for your series CPITCH, it was 20 slides of amazing business banking. And then Imad, you sheepishly showed me. I guess I have to show you by AI slide.

You could have AI slide, guys. I did do it sheepishly. I was very confident. I was like, AI slide.

Yeah, here it is. Give me the money.

I almost think founders feel like they have to do that. If you've got an amazing story, amazing customer happiness, amazing product market fit, amazing set of numbers. I couldn't care less for the AI slide. And so, and sorry. I'm very excited about what Mercury is going to do in AI.

I did it for the other VCs, Sonia, don't worry.

But I think that people think that they have to. I watch their companies and like most of us are smart enough to see through that. And at the end of the day it's like what we care about is, are you solving a real pain point? Is there a why now do you have, you know, screaming product market fit? And in those cases, we're always interested to Sarah's point, oftentimes that does correlate with whether you're grabbing an opportunity in AI, because we're very long in the tooth now for various market transitions. For the mobile transition, we're very long in the tooth. The app Store hadn't changed for a long time. We're very long in the tooth for SaaS. And so with AI, customers are now reaching for a new piece of technology for the first time in a long time. I take the developer IDE market as an example of that. Most developers don't want to switch their IDE for any reason. Now all of them are happily singing from the rooftops that they are on cursor or Windsurf. And so grabbing an AI opportunity does tend to correlate in some cases with whether you're able to find that spot spark of product market fit. But at the end of day, that's what we're looking for. Not is this an AI company?

Yeah. I would also add the bar is just way higher. I mean, it's partly because AI companies also grow so fast like this. You know, before it was very hard to find a company that grew from like 1 to 10 million in a year. Whereas that's actually like relatively common. Not that common, but still happens a lot more. So if you have those numbers, you can do it. Okay, so can I add one comment.

On this I was just coaching one of our entrepreneurs through her series a fundraising process and she just signed terms with the firm. And she was very frustrated because this is an entrepreneur who has very happy advocate customers. And she had a certain product positioning that made sense to the customers, that came from the customers. And the investors were like, we want to understand how the AI works. And Jessica didn't mean to say that out loud. The founder of this company was like, why do you people care? You do not actually understand how this works, investor friends. And she didn't mean it from a place of arrogance. She was just like, it doesn't matter, it works, right? And like, I think that both sides have some truth here. The investor audience was like, okay, well, you know, in these highly competitive markets, the storytelling for your employees and your investor base about, like, what you understand about making this AI solution works is an important ingredient to your success. And so I would say, like, thinking about, you know, whether or not your, your differentiation story really holds water and is compelling in this age is a good question. And then also I was like, okay, but like, to be clear, if your product is working with customers and your product marketing is working with customers, like, do not change that. That's the thing that is much more important. And so I think one thing that is just worth keeping in mind is you'll have different audiences and the audience that matters most is still the customers. And to Sonia's point, if you have the traction, people infer the story from the traction.

Actually, one more thing to add. I do think sometimes people think that if I want to be an AI company, I need Noam Brown Ilya Sutskever level researchers on my team. And the amazing thing about what's happened with foundation models is they very much have democratized the base and so making use of their models for your specific company. Sometimes we have companies that are technical enough where they're fine tuning and they're reinforcement tuning these models, but oftentimes it's just about that magic exists behind an apron API. And are you able to prompt chain the stuff together? Are you able to use natural language? Are you able to even vibe code your way towards a prototype that sparks something in the customer's eyes? And so I think what it means to be AI native and build with AI is the floor is way lower than people think in terms of what it means to be an AI company. You don't need to be a team of AI researchers.

Yeah, there's a good meta lesson in what Sarah said as well, that the product you're Selling to an investor is not the same as the product you're selling to customers. And I think founders find that frustrating sometimes. That's a good lesson. There's a bunch of CEOs I think the Shopify CEO recently and the Duolingo CEO are all about like, hey, we're AI company and internally you have to get on that train and use AI for everything. And it's become a bit of like a founder meme, I guess. Do you recommend some, like, what part of those practices do you recommend to people, especially like early stage stage founders in terms of how they use AI natively, internally?

So I just had Luis, one of the Duolingo founders, on me and my friend's podcast called no Priors. And so we were talking about this and if, you know, just because Duolingo is an interesting consumer app, like if you use Duolingo, the thing that they've used AI for is like they were more traditional machine learning and gamification and they've expanded the language set and the content set for Duolingo massively. Right? So like learning Spanish as a Portuguese speaker or vice versa different from as a native English speaker. And now they like support this big diversity because it's just cheaper to generate all the content. And I think that there's a set of things that are like very specific to each business and it might be content authoring for Duolingo or even like he was talking about how the animations being really good and fun for every transition in Duolingo is actually really important to engagement. I was like, okay, I cannot believe this is the thing you're thinking about. But I hear you to the things that are very common across companies. And the things that are common across companies are like, I need good financial operations, I need to be efficient with my go to market.

Right.

My engineering team should be as high velocity as possible within the constraint of quality. That makes sense in my vertical. So I kind of bucket it into these two areas where one requires creativity and aggression and one is just like, I want to be solidly on the right side of history of efficiency. So that's how I think many people are setting up their strategy where I'm going to differentiate based on the things that are very specific. And it's like, might need to build in house and then there's a bunch of operational things where we just want to be working with great partners.

I think there's almost this prisoner's dilemma if you're a big company. Like, if you don't make this manifesto and like hey guys, you have to work in an AI native way. Then you are going to get executed by entire teams of people that are operating that way. And as I look at some of the companies in our portfolio that are led by extremely technically savvy people, they fully are on board with this. And I think we're already seeing the data in terms of the number of software engineers that are being hired is changing in the valley. And I think that Dharmesh from HubSpot has this great phrase of we're going from copilots to autopilots, but now we're going to air traffic control. And I think we've gone from this copilot of the thing auto completing your line of code to autopilot of you can go tell the thing to build it for you. But I think we are increasingly moving to this air traffic control phase of AI where in part because of what these agents can now do by themselves, in part because of the amazing vertical specific agents that companies are building, there actually are agents that can do entire jobs to be done for you. And a lot of the cost of that is trending towards the cost of compute so very, very cheap. And so if you think about labor markets organizing around the thing that is scarce in the past, the thing that has been scarce has been human labor. In the future, the thing that's scarce if you have these agents running around doing things for you is the ability to have agency and creativity and taste and really act as the air traffic controller. So I very much believe that labor transformation is happening. Some of the best companies in our portfolio, I'm sure in yours, are very much operating with that mindset. And if you're a company that's not even in the co pilot phase, you're in the VS code without AI completion phase. How do you compete against a hundred person company where everybody in that 100 person is acting as an air traffic control? I think it's actually a prisoner's lemma, absolute critical necessity. Every company needs to be operating that way right now.

How many people is Mercury today? Like as a team?

Almost a thousand.

Almost one thousand.

Okay.

I think one undersold benefit of this stuff is for entrepreneurs. Many of you are probably leading teams of like the maximum scale you've ever led. I've never led more than 50 people. I don't think I'd be a very good at scale leader. I don't intend to do it but. And one of the really interesting things about AI if capabilities progress is you can have a lot of impact With a relatively small team, maybe with 1,000 people or 2,000 people, you can do what used to take 100,000 people in a more traditional setup. And so because our companies are all growth oriented, there's like very few people thinking about reducing headcount in the portfolio. But I think founders are maybe in the same place that I'm in, or many of you are, where they're like, oh, you know what, managing a couple thousand people is hard enough. Can we get away with growing without growing the base and therefore not taking the additional sort of organizational overhead of communication, recruiting and alignment and all the things that make big companies harder sometimes. And so I think having that ambition of how much more can each person do? And there was an era where people glorified building teams very large quickly. I think now there is enthusiasm that is real and deserved about, well, wouldn't it be cool if we could be a startup forever, but with several billion.

Of revenue, wouldn't that be cool? It's kind of an interesting point that AI is kind of a boon for smaller companies. One thing that probably people don't realize is even when you have more than 100 people, it's really hard to change culture. And these CEOs making these announcements are trying to do something drastic to change their culture, but it's much easier for most people in this room. With five people, you can do anything. I want to give some time for Q and A. So I guess think of some audience questions. Do we have any mics going around? Oh, there's mics. Hands up if you. There's a question here.

So, as Mercury's CEO and founder, do you feel threatened through stablecoins and how it's going to reimagine banking and also with AI? Do you believe AI with stablecoins is also going to change how people are going to send and remit money all across the world?

I mean, I see stablecoin as an opportunity, not as a threat. I think if you're doing innovative things like most change, I mean, incumbents still control 99.9% of financial services in the U.S. so most change will be good for us. We can easily adopt stablecoins and we're thinking about it. AI and stablecoins maybe. I mean, it's tricky. I do think, like, eventually we'll have native money that AI agents use. I mean, it sounds cool. We're still probably early in that. But yeah, I think the mindset of thinking of any of these things as threats is a bad mindset. Like it's all opportunity from my perspective, I think that was Jason back there.

That guy's leading the stablecoins team.

There's a question there.

Just a general question. Fundraising question. Any suggestions or recommendations for startups that are. No AI, no traction, very new market that most people don't even think that it exists.

Thank you. So what was the last point?

Most people don't even think it exists because of the new market.

Oh, new market.

I see.

I don't know if this is the answer you were looking for, but I would say probably go get some revenue or traction.

I don't.

I just, like. I think very directly like you. You have two paths. If you believe something other people don't. Right? Like, you find people who you can convince of the story. Like, look for indications. And, you know, the Internet's great. Like, there's so much content from investors now about, like, what they're thinking about. And so you can look for people who are aligned with your interests or some of your worldviews. Shock. There are VCs that have, like, some points of view independently. But the obvious thing is, go figure out whatever you can do to get customers to sign up for the vision, even if it is just being a design partner on the demand side. And good investors always listen to the customer. I think they're willing to change their worldview based on what the buyer actually says they want.

I'll also just say if it seems like you're, you know, six standard deviations above the norm in whatever field you are, and so, you know, seeing the evidence of that, and if you're able to tell a convincing story to why that matters, like, we're all ears. And so we're looking for excellence. And, like, I think the bar goes up as you go outside the strike zone. But never saying that. Like, we just invest in a company that beams sunlight down to the earth from a bunch of mirrors in space right there. Like, there's definitely no. It has to fit this underwriting box or we're not investing. And so we're just looking for excellence and many standard deviations above the average.

I have this kind of view that I've developed, which is that either you're fundable or not fundable or in the gray zone. And actually very few things in the gray zone. So it's much better to just be extremely fundable and then you'll raise money easily. So just have, like, a ton of traction and go build the thing and have customers that love it because, like, it's very not fun when you're in the Unfundable section and you're like wasting time talking to VCs because it's just all investors in general. So yeah, that's. And you can. I think sometimes entrepreneurs sit in that unfundable zone and waste a lot of time talking to investors. If you're in Silicon Valley, you can probably get a signal within a month and that would be the maximum kind of timeline I would give myself.

Hey, this has been great. I think I'm still trying to reconcile sort of two beliefs that you shared on sort of the investment side. One is that the application layer for software development coding, the foundational models are going to kind of eat that up. But on the other hand that behaviors are changing and historically you would never bet on developers to change IDEs. But now there's Windsurf, there's Cursor, there's Zed. So I'd be curious, Sonia and Sarah, to hear a little bit more about your perspective of what does the future of software development in an AI native world look like. How is that changing? Where are there opportunities? Where are there challenges? Challenges?

I don't think the application layer is going to be eaten up or I think, you know, there are some very specific markets that are going to be in and those could be very large. Like the. It is, it is possible that the search and personal assistant market, it's very related to frontier model development. ChatGPT. Right. It is possible that next line code generation or um, like you know, code agent command is something that is so core to model progression. There's a. This is an extremely mainstream theory around the found in the foundation model labs that improving code reasoning will generalize to other types of reasoning and we should do a lot of it because we can generate and validate code. Right. Um, so there are areas that matter a lot to the labs. I think these are the two. I think. And by the way we have companies that compete in software engineering because for anybody who has worked on a software engineering team, the job is not just next line completion. Right. And we are still far away from the models being able to understand a customer need, figure out what the implementation should be, understand the dependencies, collaborate with the designer and be like, I actually don't think the is going to work and fit it into this sprint with the 16 other bugs that they were supposed to deal with of varying importance. I think the job is just much more complex than that. I think we are going to get a set of tasks that will be agentic, I think will be faster in that actual writing of New feature development. The tasks that might be automated include things like debugging. We should have better log parsing, automated security analysis. When you commit code, there's lots of things that we should benefit from, but I do not. As far as I can tell, from now until the visible future, we're hiring more engineers in our companies, not less. So that's what I say.

But maybe don't do another VS code fork. But I think we're at time. Can we do another question or are we time? One more question. All right. Whoever gets the mic put in front of them will get the last question.

All right.

There.

I'm honored. Oh, that's my voice too. Okay, so first thing, quick thing. If you guys have any more of those Mercury jackets, my size is a 2 XL. I just wanted to let you know I saw them, they look nice. So good job on that. But my questions specifically is I feel like I'm probably the only non profit person in here. I wanted to know what you guys thought, how AI could like or how nonprofit organizations could benefit from AI and what you guys usually invest in. Not asking for investment, but just it kind of got my brain like thinking, like, what are some things that my nonprofit can benefit from, like utilizing more AI and investing in AI in our organization.

What does your nonprofit do?

So we provide after school activities to low income kids in Southern California.

Nice.

It's called Our Way Home. Thank you. I'll take that applause. But my wife was the main founder. I'll give it to her.

I'll start with a couple thoughts. So one that is really cool. So thank you for sharing what you do. I think of AI as the great democratizer of opportunity. And there's still just this gap to come up, that behavioral curve on how do I make the most of this technology. I think all of us are still even figuring out how to make the most out of ChatGPT, right? And 6 billion people on the planet haven't tried it yet. And so to me, because AI is the great equalizer and because it is such an opportunity, providing that last mile and getting people to. Getting people to make the most of it, I think is really powerful. So for example, in education, I think there's Bloom's two sigma principle and the idea that if you give every kid a personalized tutor that they can perform two standard deviations better, I fundamentally think every child deserves right to an education at that level. And even though the technology now exists, it's not being brought to everybody. And so I think there are plenty of opportunities like that including education, where I think, you know, you could actually your nonprofit could be driving that last mile adoption. And I think that's extremely powerful for people. It's kind of to Sarah's point on, like, you can use AI to either change your product or to make yourself operationally more efficient. And like I would say for a nonprofit, you can make yourself operationally more efficient with AI, I believe that opportunity exists. And making yourself operationally more efficient, that's more dollars that go to the cause. Right. But I think the more interesting thing is how can you actually change the cause and the impact you're having? And I think this is where AI and education specifically is going to be going to be a huge difference.

Can I very briefly to this. Yeah. So it is an amazing mission, I think one. So my husband and I chaired something called the Tipping Point Gala this past year. They help a bunch of San Francisco nonprofits. And one of the things they do is like one of your core jobs as nonprofit is raise money, right? So you can go deploy your programs. One of the things they do is help nonprofits basically do analytics on their outcomes and tell their stories. These, I guess they're two things so that they can raise more money. Right. And this is just a core function. I actually think AI is going to be really helpful for both of these things. Right. So you can use like free personal AI tools like ChatGPT or Cloud or whatever to take a bunch of data and ask questions of it that are structured. Instead of paying somebody who knows SQL or building a data warehouse, you just dump the data in and get pretty useful answers out. And on the storytelling piece, I think that the ability to use really natural interfaces like record video of your kids telling stories about what's going on with their activities and then sharing that is one thing, but then also sharing the aggregates and being able to pull quotes and things out with AI, I think is also going to be really compelling. And so maybe like, to me, everything looks like fundraising, but I think these pieces of fundraising will be helped.

All right, I think we could talk for another hour, but we got to wrap it up. Thanks for joining us, Sonia and Sarah.

Thank you, Sam.